Streamline scripts

[andrew-MET]

I haven't tested this at all, so there could be something that doesn't quite work. Let me know.

Main changes

• use names(thresholds) to get parameters
• use switch (R's version of case) to get parameter specific lead times and scalings
• use first_validdate and last_validdate to set the dates needed for observations for TEMP script
• Make specification of FCST_MODELS in bash script a bit simpler - now just a comma separated string
• Don't need harpIO for SYNOP script. Don't need proj4 module at run time (only compile time)
• Used automated code formatting so it looks like a lot more has changed than actually has!